04-24-2014 05:07 AM
Hi
This is Ramesh from MaxEye Technologies.
I am new to VideoMaster and PQA. Can you please let me know how to detect unwanted screen while changing channels in Set-Top-Box.
Regards,
Ramesh
04-24-2014 08:54 AM
Greetings Ramesh! Can you better define what you are looking for in your "unwanted screen"? Are you looking for defects? A scrambled picture? No picture at all? A correct picture, but the wrong channel? If you can help better define what you are looking, one of our Applications Engineers should be able to follow up and help you get PQA set up. PQA will be the correct solution for this, VMS (formerly VideoMASTER) is more for analyzing static pictures and frames. Reviewing this article may help you explain better what you are looking for.
Regards,
04-24-2014 09:37 AM
Thanks Paul.
The following are some tests we need to do
1. We need to detect the following during channel change
i. Blank Screen (No Picture)
ii. Distorted or Scramble Picture
iii. Audio distortion
2. Occurance of freezing/distorion when one channel is played continuously
3. Particular string display at random location (finger printing)
4. Audio level testing on the right and left channel
I hope all the above tests can be done with the PQA and AudioMaster.
Regards,
Ramesh
MaxEye Technologies, Bangalore
04-25-2014 10:01 AM
HI Ramesh,
In accordance to our previous conversation and owing to the general benefit, I will re iterate the points of our discussion as the reply to this post.
All of these can be achieved with PQA and necessarily need a standard test video which can be programatically generated using NI LabVIEW and VDM. This is required as we cannot afford to have 2 unknown parameters to solve an equation.
1. We need to detect the following during channel change
i. Blank Screen (No Picture)
ii. Distorted or Scramble Picture
iii. Audio distortion
To measure the balnk screen an test video stream which has a alternating squence of two frames with two standard shapes each in a contrasting colour. Noiw the absence of the shapes in the frame is a direct indication of the blank screen and deformity in shape is the distortion of the picture which can simply be measured with PSNR or a basic particle analysis on the image. (User processer required for the second method)
To measure the audio distortion we need to have a single tone or a standard and known multitone and the FFT on it will give us the amount of distortion on it and the noise floor. Audio mAster provides a very good level of analysis options for that.
2. Occurance of freezing/distorion when one channel is played continuously
A test video with frames having their frame numbers as the only element in the frame is used. Using ocr to read the numbers on them and validate for the increment. The moment there is no increment of numbers and the frame repeats then that particular fram is a filler frame - freezing. (User processor needed)
3. Particular string display at random location (finger printing)
Using OCR and some basic image processing and identification techniques this can be located and read. (User processor needed)
4. Audio level testing on the right and left channel
We can use the Audio SPL option in the NI PQA processors to process and validate it. This also requires a standard monotone to be played as a input signal.
Hope this helps. Do revert for further queries. (You can email me or call me )
Regards,
N A Gokul
04-25-2014 11:09 AM - edited 04-25-2014 11:09 AM
Ramesh, I contacted our India support and systems team, and understand that you have also begun working with them. I think it will be best to continue to work with them on the lower level implementation details where you can have more direct contact and work in your own time zone, but I will be glad to help answer some of your questions here, as well.
Before I get to the below explanations, I would like to point out that some of your testing requirements are going to be difficult without a reference stream, or a known video source. Without being able to control the content, in some cases it is going to be difficult to know what we are looking for. The quick explanation is that the methods for doing these non-referenced measurements are still being researched by experts in the field. There are some proposals, algorithms and papers out there on the topic, but as far as we are aware this hasn't been implemented in the real world yet.
Additionally, as I was writing this Gokul added in his comments (Thanks Gokul!), so I will try to build on those with some additional thoughts and details.
As for your tests, here are my initial recommendations:
1. We need to detect the following during channel change
i. Blank Screen (No Picture)
ii. Distorted or Scramble Pictue - This is going to be the tricky one because if we don't have a known video source, a reference, to compare to, its difficult to know when a picture is bad
iii. Audio distortion:
2. Occurance of freezing/distorion when one channel is played continuously
3. Particular string display at random location (finger printing)
4. Audio level testing on the right and left channel
One other general note. We are talking about a lot of processing here, and possibly the need for some post processing. Depending on the power of your PC and the video resolution, it may not be able to perform all of these calculation on-line, or on the fly. We may also need to look at performing some of this in post processing.
04-25-2014 11:34 AM
Thank Paul and Gokul for the detailed reply.